Goals And Achievements After One Year

SFX Knowledge Base Advisory Board (KBAB)

François Renaville, University of Liege

Mark Needleman, Florida Virtual Campus

IGeLU 2014 Conference Oxford

September 15-17, 2014

Download: http://hdl.handle.net/2268/171986

Abstract

Founded in 2013 as a result of discussions at the 2012 Zurich

Conference, the goal of the SFX Knowledge Base Advisory Board

(KBAB) is to promote first class quality of the data stored in the

SFX Knowledge Base by reviewing the quality assurance policies

and processes together with Ex Libris. During its first year, the

group identified several issues in order to improve the CKB

quality. They shared them with Ex Libris which agreed to work on

or to take a look at some. This session will explain how KBAB has

been working, and present some of the issues and the

improvements brought by Ex Libris.

2

SFX

Kn

ow

led

ge

Ba

se A

dvi

sory

Bo

ard

(K

BA

B)

Goals

• Founded in 2013 as a result of discussions at the 2012 Zurich

conference

• Goal: to promote first class quality of the data stored in the

SFX Knowledge Base by reviewing the quality assurance

policies and processes together with Ex Libris

• In the long run the SFX KBAB is supposed to:

• Create its own ideas for new KB policies and/or processes and

propose them to Ex Libris

• Receive, review, and forward to Ex Libris any such ideas

submitted from the SFX user community

• Provide feedback and advice on intended extensions or changes

of KB policies and/or processes on according requests from Ex

Libris.

3

SFX

Kn

ow

led

ge

Ba

se A

dvi

sory

Bo

ard

(K

BA

B)

Working Body

The SFX KBAB is a joint IGeLU and ELUNA group.

• IGeLU

• François Renaville, University of Liege Library, Belgium, Coordinator

• Mark Needleman, Florida Virtual Campus, USA, Deputy Coordinator

• Yosef Branse, University of Haifa Library, Israel

• Holly Thomason, Stanford University, USA, liaison IGeLU SFX PWG

• ELUNA

• Stephanie Nicely Aken, University of Kentucky, USA

• Erika Banski, University of Alberta, Canada

• Xiaotian Chen, Bradley University, USA

• Ann Ercelawn, Vanderbilt University, USA

• Marina Oliver, Texas Tech University, USA, liaison ELUNA SFX PWG

• Contact with Christine Stohn, SFX Product Manager 4

SFX

Kn

ow

led

ge

Ba

se A

dvi

sory

Bo

ard

(K

BA

B)

Under the Microscope

• +/- 35 discussed issues (9 in a priority list) during the year

• Related to CKB

• Sometimes software issues -> out of scope -> frustrating

• Focus on some issues:

1. E-books metadata

2. SFX subject categories

3. Undef

4. Individual volume names for monographic series

5. Initial articles at the end of titles

6. Language + initial article

7. Beginning and ending dates

8. MISCELLANEOUS_FREE_EJOURNALS

5

SFX

Kn

ow

led

ge

Ba

se A

dvi

sory

Bo

ard

(K

BA

B)

(1) E-books metadata

• More complete metadata

• E-books authors are too often missing

• Categories are missing

• Publication date

• Subtitles

• …

• Encouraging vendors to enhance metadata

• By Ex Libris

• By SFX Community

• Possible to come up with a letter to vendors?

6

SFX

Kn

ow

led

ge

Ba

se A

dvi

sory

Bo

ard

(K

BA

B)

Ex Libris’ feedback

• KBART recommendations Phase II include the addition of

author information for E-Books.

• However, it is still often missing or the information is very

difficult to manage and inconsistent.

� Ex Libris promotes the use of KBART II

recommendations to content providers.

• Also investigating if ExL can enrich the data from other places

(but this is a longer project).

7

SFX

Kn

ow

led

ge

Ba

se A

dvi

sory

Bo

ard

(K

BA

B)

(2) SFX subject categories

• SFX subject categories are not systematically added to objects

• Known problem for e-books (see above)

• But also concerns many e-journals

• Next to the quantity of categories, category assignation should

be logical, relevant and accurate (quality aspect).

8

SFX

Kn

ow

led

ge

Ba

se A

dvi

sory

Bo

ard

(K

BA

B)

Analysis 1

• University of Haifa Library (January 2014)

• 146,185 active object portfolios (journals) -> 97,043 distinct

objects after de-duplication.

• 56,771 (58.4%) have no category at all

9

SFX

Kn

ow

led

ge

Ba

se A

dvi

sory

Bo

ard

(K

BA

B)

82,4%

6,2%

2,7%

1,9%1,9%

0,7%0,5% 0,4% 3,4%

English [n=46,753]

Japanese [n=3,540]

Spanish [n=1,520]

French [n=1,054]

German [n=1,053]

Portuguese [n=406]

Hebrew [n=265]

Russian [n=240]

Others [n=1,940]

Thanks to Yosef Branse!

Analysis 2

• University of Liege Library (January 2014)

• 81,881 active object portfolios (journals) -> 62,769 distinct

objects after de-duplication.

• 34,449 (54.9%) have no category at all

10

SFX

Kn

ow

led

ge

Ba

se A

dvi

sory

Bo

ard

(K

BA

B)

90,9%

2,9%2,8% 2,3% 1,0% 0,0%

English [n=28,944]

French [n=926]

Spanish [n=900]

German [n=729]

Portuguese [n=332]

Others [n=2618]

From those 34,449 objects with no category:

• ISSN?

• 17,494 (50.8%) have got a print ISSN

• 5,405 (15.7%) have got a online ISSN

• 4,615 (13.4%) have both

• �Almost 65% have got at least one ISSN

• Publishers?

• Elsevier: 262

• Springer: 249

• Taylor & Francis: 55

• Wiley: 49

• SAGE Publications: 36

• NPG: 1511

SFX

Kn

ow

led

ge

Ba

se A

dvi

sory

Bo

ard

(K

BA

B)

Ex Libris’s feedback

• Ex Libris is testing options to add substantially more categories

by using CONSER as the basis and map the categories to SFX.

• If successful � can be done on an ongoing basis.

• Mapping table and process are created

• Ex Libris is in the midst of testing the result (a few more weeks)

and might have to adjust the process several times.

• No plan to add subject categories to e-books – given the large

number of e-books as opposed to e-journals

12

SFX

Kn

ow

led

ge

Ba

se A

dvi

sory

Bo

ard

(K

BA

B)

(3) Undef

• Thresholds should avoid "undef" when possible

• Some journals start or end full text in the middle of a year, and

targets vendors may have clearly stated that.

• Eg: Journal of Nursing Regulation on CINAHL.

• EBSCO says on CINAHL that the full text coverage is "10/01/2010 to

present" (October 2010-present)

• but SFX KB says "$obj->parsedDate('>=',2010,undef,undef)".

� Dead links happen when earlier 2010 issues are needed.

• Double ‘undef’ is especially source of trouble

� Analysis of samples from SFX 4 Revision 20143200

13

SFX

Kn

ow

led

ge

Ba

se A

dvi

sory

Bo

ard

(K

BA

B)

Double ‘Undef’ Publishers Targets

getFullTxt Targets

portfolios

with

(e)ISSN

At least 2 ‘undef’

(for volume and issue) in

THRESHOLD_GLOBAL

AMERICAN_CHEMICAL_SOCIETY_JOURNALS 75 0 0%

ANNUAL_REVIEWS_COMPLETE 58 0 0%

BRILLONLINE_JOURNALS 252 0 0%

CAMBRIDGE_UNIVERSITY_PRESS_JOURNALS

_COMPLETE469 0 0%

ELSEVIER_SD_SCIENCE_DIRECT_COMPLETE 3,698 24 0,6%

EMERALD_EJOURNALS_PREMIER 350 1 0,3%

OXFORD_UNIVERSITY_PRESS_COMPLETE 324 14 4,3%

SAGE_COMPLETE 803 26 3,2%

SPRINGER_LINK_JOURNALS_STANDARD 2,983 1,107 37,1%

SPRINGER_LINK_ONLINE_JOURNALS_ARCHI

VE_COMPLETE1,251 223 17,8%

TAYLOR_FRANCIS_ONLINE_COMPLETE 2,457 76 3,1%

WILEY_ONLINE_LIBRARY_JOURNALS 2,490 85 3,4%

14

SFX

Kn

ow

led

ge

Ba

se A

dvi

sory

Bo

ard

(K

BA

B)

Double ‘Undef’ in Third Parties Targets

getFullTxt Targets

portfolios

with

(e)ISSN

At least 2 ‘undef’

(for volume and issue) in

THRESHOLD_GLOBAL

CAIRN_GENERAL 425 10 2,4%

HIGHWIRE_PRESS_JOURNALS 1,647 1,596 96,9%

INGENTA_CONNECT_JOURNALS 7,797 611 7,8%

JSTOR_ARTS_AND_SCIENCES I -> XIII 2,473 103 4,2%

METAPRESS_JOURNALS 2,140 3 0,1%

OVID_JOURNALS_AT_OVID 3,020 7 0,2%

PROJECT_MUSE_STANDARD_COLLECTIO

N343 6 1,7%

15

SFX

Kn

ow

led

ge

Ba

se A

dvi

sory

Bo

ard

(K

BA

B)

Double ‘Undef’ in AggregatorsTargets

getFullTxt Targetsportfolios

with (e)ISSN

At least 2 ‘undef’

(for volume and issue) in

THRESHOLD_GLOBAL

EBSCOHOST_ACADEMIC_SEARCH_COMPLETE 6,762 6,741 99,7%

EBSCOHOST_ART_ARCHITECTURE_COMPLETE 349 348 99,7%

EBSCOHOST_BUSINESS_SOURCE_COMPLETE 3,856 3,847 99,8%

EBSCOHOST_COMM_MASS_MEDIA_COMPLE

TE497 494 99,4%

GALEGROUP_ACADEMIC_ONEFILE 5,515 5,496 99,7%

GALEGROUP_GENERAL_ONEFILE 8,482 8,466 99,8%

PROQUEST_ABI_INFORM_COMPLETE_NEW_P

LATFORM4,300 4,287 99,7%

PROQUEST_ENVIRONMENTAL_SCIENCE_COLL

ECTION_NEW_PLATFORM1,093 1,091 99,8%

PROQUEST_EDUCATION_COMPLETE_NEW_PL

ATFORM1,027 1,026 99,9%

PROQUEST_CENTRAL_NEW_PLATFORM 5,791 5,765 99,6%16

SFX

Kn

ow

led

ge

Ba

se A

dvi

sory

Bo

ard

(K

BA

B)

Ex Libris’ feedback

• Ex Libris KB team does give this issue special attention

• Verifies with vendors if this information is missing but

obviously they need their cooperation.

• Alternatives under analysis:

• Conversion of date information into volume/issue.

• Difficult to do if there is no indication of what issue was published in

what month.

• ExL is looking at options (not a quick fix!)

• Also considering:

• Rule that checks if there already is a volume and issue in the KB,

• If the year is the same � ‘undef’ overwritten with it.

• !! May have the implication that volume/ issue is wrong!! 17

SFX

Kn

ow

led

ge

Ba

se A

dvi

sory

Bo

ard

(K

BA

B)

(4) Individual volume names for

monographic series

• Many monographic series have individual titles for volumes

• Problem:

• Library may not subscribe to the entire series

• Even if entire series is subscribed to user may only have the

title of an individual volume

• Need invidual titles and ISBNs in KB for these

• Some progress has been made in this area – more work needs

to be done

18

SFX

Kn

ow

led

ge

Ba

se A

dvi

sory

Bo

ard

(K

BA

B)

Ex Libris’ feedback

• Ex Libris does add those, and specifically asks vendors to

provide them if they are missing.

19

SFX

Kn

ow

led

ge

Ba

se A

dvi

sory

Bo

ard

(K

BA

B)

(5) Initial articles at the end of titles

• Problem:

• Some publishers put the initial article at the end of the title

• Nice to have more standard practice on handling this

• Possible to have some sort of Normalization done when loaded

into KB?

20

SFX

Kn

ow

led

ge

Ba

se A

dvi

sory

Bo

ard

(K

BA

B)

(6) Language + initial article

• Problem:

• Many cases of initial articles in non-English languages not being skipped

• Problem with journals being in one language but the language in KB is

different

� Causes problems with searching and with facets in Primo

� Impact on users & their perception of Primo

• Batch processes when loading/updating CKB possible?

• Must be done carefully!

21

SFX

Kn

ow

led

ge

Ba

se A

dvi

sory

Bo

ard

(K

BA

B)

22 SFX Knowledge Base Advisory Board (KBAB)

(7) Beginning and ending dates

< came late under discussion >

• ISSN Register has beginning and ending dates

• This information should be added into KB:

• Would help avoid portfolios being attached to the wrong title

when journals have the same title

• Would also help with creation of thresholds or portfolios

23

SFX

Kn

ow

led

ge

Ba

se A

dvi

sory

Bo

ard

(K

BA

B)

Ex Libris’ feedback

• Ex Libris:

“This is out of scope for the time being I’m afraid. We can

get back to it later but it also requires a technical change

in the database, it’s not just about KB data.“

24

SFX

Kn

ow

led

ge

Ba

se A

dvi

sory

Bo

ard

(K

BA

B)

(8) MISCELLANEOUS_FREE_EJOURNALS

• Many problems

• Not all journals are free

• Incorrect thresholds

• Bad or outdated parse params

• Analysis of a sample of the 24,236 portfolios (April 2014)

• Sample of 2% (484 portfolios)

• 96 cases (19,8%) with an incorrect parse param (like 404 File Not

Found, no journal page anymore, no platform anymore).

• 388 cases (80,2%) with a correct parse param: linking to the home

page of the journal, to the last issue, to the archive or to a search

form

(Thanks to Myriam Bastin)

• Duplication with titles in other free targets

• A lot of work to maintain25

SFX

Kn

ow

led

ge

Ba

se A

dvi

sory

Bo

ard

(K

BA

B)

• KBAB carried out a survey among SFX Users

• About 120 responses

• Most respondents do use the target

• Slight majority do selective activation

• About half do Auto-Activate

• Full analysis available in autumn 2014

• Some desired changes from Ex Libris:

• More sytematic checking of thresholds and Parse Params

• Creating a new Target for Miscellaneous Academic Journals

• Also suggested: new subtargets per language

• Deleting journals that exist in another Free or Open Access Target

• Getting the CONTRIBUTE button more used by the clients (promotion…) 26

SFX

Kn

ow

led

ge

Ba

se A

dvi

sory

Bo

ard

(K

BA

B)

Conclusions

• Very long-term task (if not endless)

• More work could have been done, but

• Also a time-consuming task for KBAB volunteers & KB Team

• Sometimes frustrating when KB quality and metadata issues are so

closely linked with software issues

• For Ex Libris, “the priorities help the KB team to specifically look for

things and change their processes accordingly. So even if there is no

big leap forward, this is very helpful”.

• For the future

• So far, KBAB has worked with few outside contact

• Should certainly start to open to the SFX community (without

getting bogged down in hundreds of requests!)

• Suitable way to work has to be found!!

27

SFX

Kn

ow

led

ge

Ba

se A

dvi

sory

Bo

ard

(K

BA

B)

28

SFX

Kn

ow

led

ge

Ba

se A

dvi

sory

Bo

ard

(K

BA

B)

francois.renaville@ulg.ac.be | mneedlem@ufl.edu

http://hdl.handle.net/2268/171986